Free, source-traceable HIV knowledge

Confidence without false certainty

Evidence Classification

The Center's multidimensional evidence states, from established guidance to preliminary, conflicting, refuted or retracted findings.

Publisher
The Bach Foundation
Audience
Public · Professional · Research
Last reviewed
18 August 2026
Status
Published standard
Jurisdiction
Global context; local guidance may vary

Commitment at a glance

No single badge can summarize scientific truth. Evidence labels always sit beside the source type, population, outcome, date, limitations and jurisdiction.

Classification standard reviewed 18 August 2026

Clinical and public-health states

  • Established: supported by converging evidence and accepted practice within a defined scope.
  • Guideline supported: recommended by a named authority for a stated population and jurisdiction.
  • Strong or moderate evidence: confidence reflects design, consistency, precision, directness and risk of bias.
  • Implementation evidence: effectiveness depends materially on setting, delivery capacity and access.

Research states

  • Clinical research: being evaluated in people; no implication of approval or effectiveness.
  • Early clinical: primarily safety, feasibility, dose or biological-activity evidence.
  • Preclinical: laboratory or animal evidence that cannot establish human benefit.
  • Promising but unproven: credible signal with important unanswered questions.

Integrity states

  • Conflicting or inconclusive: available results do not support one stable interpretation.
  • Not supported or refuted: evidence fails to support, or positively contradicts, the claim.
  • Corrected: the source remains available with a material amendment.
  • Withdrawn or retracted: the integrity notice remains visible and linked.

What labels never mean

Labels are not product endorsements, treatment recommendations, numerical truth scores or substitutes for reading the source. They cannot erase population differences or convert an association into causation.